Bridging the Projection Gap: Overcoming Projection Bias Through Parameterized Distance Learning

Chong Zhang, Mingyu Jin, Qinkai Yu, Haochen Xue, Shreyank N Gowda, Xiaobo Jin*

*Corresponding author for this work

Research output: Chapter in Book or Report/Conference proceedingConference Proceedingpeer-review

4 Downloads (Pure)

Abstract

Generalized zero-shot learning (GZSL) aims to recognize samples from both seen and unseen classes using only seen class samples for training. However, GZSL methods are prone to bias towards seen classes during inference due to the projection function being learned from seen classes. Most methods focus on learning an accurate projection, but bias in the projection is inevitable. We address this projection bias by proposing to learn a parameterized Mahalanobis distance metric for robust inference. Our key insight is that the distance computation during inference is critical, even with a biased projection. We make two main contributions - (1) We extend the VAEGAN (Variational Autoencoder \& Generative Adversarial Networks) architecture with two branches to separately output the projection of samples from seen and unseen classes, enabling more robust distance learning. (2) We introduce a novel loss function to optimize the Mahalanobis distance representation and reduce projection bias. Extensive experiments on four datasets show that our approach outperforms state-of-the-art GZSL techniques with improvements of up to 3.5 \% on the harmonic mean metric.
Original languageEnglish
Title of host publicationAsian Conference on Computer Vision (ACCV) 2024
Place of PublicationACCV 2024
Pages3327-3343
Volume15472
Edition1
ISBN (Electronic)1611-3349
DOIs
Publication statusPublished - Sept 2024

Keywords

  • Generalized zero-shot learning
  • Mahalanobis distance
  • Projection bias

Fingerprint

Dive into the research topics of 'Bridging the Projection Gap: Overcoming Projection Bias Through Parameterized Distance Learning'. Together they form a unique fingerprint.

Cite this